紀要論文 GPU を用いたマルチスレッドアルゴリズムの 効率的な実現に関する一考察

岡田, 圭太

内容記述
A GPU is generally an arithmetic unit specializing in image processing. Although its function is limited as compared with CPU, a large amount of data can be simultaneously and concurrently processed by a plurality of processors. In recent years, attention has been drawn to GPGPU which makes use of this GPU function not only for image processing but also its processing capability for high performancecomputing.In this thesis, we can consider efficient implementation of multithread algorithms by using Dynamic Parallelism in GPU. We give efficient algorithms to Fibonacci sequence and Tower of Hanoi. In particular, our algorithm for Tower of Hanoi is better than previous algorithm.Key Words : GPGPU, multi-thread algorithm, Dynamic Parallelism(DP),
本文を読む

http://repo.lib.hosei.ac.jp/bitstream/10114/13565/1/15R4109.pdf

このアイテムのアクセス数:  回

その他の情報